Abstract

The objective of this project is to analyse the housing market in Ireland from 2010 to 2019 using strong visualisations to support our analysis. We will also offer house price predictions for 2020 through to 2022. In addtion we will conduct a more granular study of commuter and costal towns offering analysis of trends relating to this data. In the next section we will provide a general overview starting with a brief discussion about the recession, its effects and the aftermath. This will be followed by a look at changes in house prices over time at a regional level in Ireland.

Overview

The global economic crash of 2008 had a profound effect on the housing market in Ireland.[2] Ireland was severly hit becomming the first european country to offically enter recession status in september of 2008 and which lasted well into 2013 and which effectively brought to an end to what is colloquially refered to as the ‘Celtic Tiger’, a period of economic growth that originated in the early 1990’s.

One of the primary reasons the economy was so deeply effected was due to a property bubble which arose due to poor financial regulatory practices relating to lending and mortage policies by banks.[3]

To aid in our overview a Choropleth animation was created using data from the Ordinance Survey Ireland group.[4] The chloropleth animation shows the median property prices for each county for each year ranging from 2010 to 2019 (Fig. 1). A property in this context can mostly be defined as either an apartment or a house. Starting in 2010 we can see a continued decline in median property prices country wide up until 2013. At this point property prices for 11 counties had a median price of 90K or less. After 2013 propety prices in general begin to recover increasing year on year for most counties. A point of note is Dublin which has the greatest increase in median price and from 2017 on the surrounding counties of Meath, Kildare and Wicklow had median prices above 210K. By the end of 2019 Galway and Cork also went above the median price of 210K.

In general property prices appear to be increasing and further detail will be given in the following sections

Data Cleansing

The dataset used in this analysis was downloaded on the 03/04/2020 as a .csv file from the Property Services Regulatory Authority website [1]. At time of download it contained 416380 observations with nine variables. Before data analysis commenced, the dataset was read-in and examined. The price variable required removal of the euro symbol and commas, and new properties had their prices adjusted to include VAT (stored in a new variable ‘price_adj’). Redundant variables were removed and a new variable containing just the year of sale was created (‘year’). Property addresses and county names were converted to lower case and non-alphanumerical and non-punctuation characters were removed. The dataset was checked for missing values (none present) and outliers. There were a number of very low priced properties - we did not treat these as outliers as they could be either very rundown properties, bedsits above a shop, or a very small plot of land (particularly during the recession when prices were at an all-time low). Higher priced observations were looked at in more detail, with the majority identified as multi-unit properties. The final step in the data cleansing section involved splitting these properties into separate observations, each imputed with the mean adjusted price for that address. Observations greater than 10 million euros were removed as we felt that these most likely represented multi-unit properties rather than an individual property. There are a number of duplicates in the dataset, however some are genuine (the same property sold more than once), some are present as an artefact of splitting multi-unit properties, with the remainder as either genuine errors or multiple properties sold together under the one address. As there was no way of determining which were errors and which were genuine, we left them in. We felt the time cost of trying to extract the genuine errors from the data outbalanced the value of doing it. The cleaned dataset was saved as a .csv and named ‘PPR-ALL-CLEAN.csv’.

Property price change by region

For this analysis, Ireland was divided into nine geographical regions. Four towns per region were selected: a primary town (urban/large), followed by three more towns with increasing distance from the primary. The aim was two-fold: investigate whether property prices change with increasing distance from the primary town, and determine whether a regional difference in property prices exists.

Fig. 2: Property price change for towns within different regions of Ireland (2010-2020)

In general, primary towns, and moreso, towns within 10 km of these, tended to have the highest median property prices across all years and regions (Fig. 2). Noticeable exceptions include Maynooth (Eastern) and Slane (North East), where both exceeded the median property price of the primary and it’s next closest town, for at least half of the time period. This could be due to the fact that the proportion of houses to apartments increases further out from urban areas, with houses typically selling at a higher price than apartments. Both towns are also popular commuter towns, with demand pushing up prices in these areas. The North showed very little differentiation between it’s four towns. The Northern region, along with the Midlands and North West (excluding Strandhill), had the lowest median property prices of all the regions; the Eastern region had the highest. Most towns saw a drop in median property prices from 2010 to 2012/2013, with the majority seeing a recovery and upwards trend from 2014 onwards.

Analysis of Dublin Commuter Belt

The Dublin commuter belt was originally an area of land beyond the M50 motorway where people who worked in the city would buy homes which were more affordable than the properties in the inner city. The Dublin commuter belt initially comprised towns such as Lucan and Swords. However, since the height of the housing bubble in the early 2000’s the Dublin commuter belt began to spread into the adjoining counties of Kildare and Meath until it’s halt was eventually precipitated by the 2008 housing collapse. The heat map below shows the median house and apartment prices in the period from 2010 to 2019 in the counties of Kildare and Meath. The heat map is illustrative of the sharp fall in property prices that was witnessed in the years from 2011 until 2013 followed by a resurgence in property prices from mid 2014 onwards. Property prices grew strongly from 2014 where the median price was €193,000 until the end of 2019 where the median price had risen to €268,722 in the Kildare, Meath areas.[5]

Comparision of New properties vs Second-hand properties

When it comes to house buying there is a huge dilemma whether to buy a new property or a second-hand property and whether to buy it in Dublin towns or other towns. The below interactive plot helps us in making this decision by understanding the trends across the last ten years.

The time-period between Q3 2012 and Q1 2014 marked in red in plot 1 is the lowest point of prices for both new and second-hand properties for the dublin towns. One the other hand, the time-period between Q1 2013 and Q2 2014 marked in red in plot 2 is the lowest point of the prices for both the new and second-hand properties for the other towns excluding dublin. While the new and second-hand properties both see their highest points in Q1 2020 marked in green in both the plots we see a huge difference in the trend of second-hand properties between dublin and other towns.

After the Q1 2014, the second-hand properties of other towns did not increase that much while the second-had properties in dublin shows an increasing trend. This shows us that the second hand prices trend will go even higher and someone who’s trying to make an investment in the second-hand properties should prefer buying in the dublin towns compared to the other towns.

On the other hand, buying a new house/apartment will be profitable both in dublin and other towns on the long run. We see a steady growth in trend of the new house/apartment prices for other towns but in dublin we see a sudden decreasing trend in Q2 2015 which again picked up in the next quarter.

This is the trend analysis that we can see considering the limitation that we do not have information regarding the size of the property and will be impacted by the number of sales in each quarter too.

In the below graph, one can zoom in to any specific time for the last 10 years and compare it’s new properties vs second-hand properties trend between dublin and other towns at the same time as the two plots are synchronized too for easy access.


References

[1] https://www.propertypriceregister.ie/website/npsra/pprweb.nsf/page/ppr-home-en
[2] https://en.wikipedia.org/wiki/Post-2008_Irish_economic_downturn
[3] https://en.wikipedia.org/wiki/Celtic_Tiger
[4] https://data.gov.ie/dataset/counties-osi-national-statutory-boundaries-generalised-20m/resource/b412ae22-ea13-4ca3-a8be-5ffc21f455f6
[5] https://www.irishtimes.com/business/economy/priced-out-home-buyers-drifting-to-dublin-s-commuter-belt-1.4104686